>
> To clarify, the index is based on a function called "split_part(....)
> The WHERE clause is only referencing the full_part column, so the planner
> cannot associate the index with the full_part column.
>
Thanks for the explanation, Melvin.
It would be simple like:
CREATE INDEX CONCURRENTLY ON gorfs.inode_segments USING btree ("full_path");
?
Thanks again.
Lucas